Why I don’t let AI write for me.

I am a writer.

My latest book, Far Horizons, took me ten years.

I grappled with every word.

Changed the tense of the entire book from past to present and then back to past, which means that I had to rework EVERY sentence. Twice.

When I felt like a bit was done, I would print it out, and then cover it with pen marks over the next hour as I sat in a chair and proofed. Chapters that I thought were brilliant, I tore up. I gave myself chills writing some pieces. Sometimes I cried. Sometimes I smiled from ear to ear. Thankfully, thousands have now read the book, and many have reached out telling me of their own tears or joys as they did. Of their wishes that there were more pages to sink into or for a sequel. Even without these moments of connection, if I’d never sold a copy but could go back to the beginning, I would do it all again. Because this little thing I’m preciously calling art was first of all created for me, the artist. That others have found beauty in it is just a bonus.

Last year, I was curious about how well AI could write. I played with Anthropic’s incredibly impressive Claude AI and found that with a little prompting, it could write very well. So well, in fact, that I decided that I would make a commitment to write without it.

I want to be deeply human in my writing. I want to deliberate over every word. Sometimes I want to dig down into my soul. Often, I want to create. I hope to one day write something that Ocean Vuong describes as “a new contribution to the species”.

I don’t even let Grammarly or Microsoft Word homogenise my sentences.

Nick Cave just wrote about the AI songs that are coming.

“Before long, they will be able to produce songs indistinguishable from those created by humans. And this is what grieves me. They will be identical in presentation, perhaps even superior, but entirely devoid of soul, cynically undermining the need for matters of the spirit, the sacred, the divine.”

It’s currently very easy to spot AI writing, and this month it seems to be popping up on every second post on Instagram or LinkedIN. The easiest giveaway is the mechanism of “its not ___, its ___.” So, for example, “it’s not just this season’s trend, it’s a revolution.” I get that not everyone wants to write like I do, but I think if you are pressing send on an AI written piece that you think is heartfelt and profound, many people may be reading it and sensing that it is anything but. If they read it at all beyond the first “it’s not ___, it’s ___.” I’d much prefer to read something real and raw and written by you.

This post took me ninety minutes. I wrote it after the kids went to bed. Checked it three times. Read it out loud twice. All the mistakes are mine. AI could have spat it out in 12 seconds. I’m OK with my personal investment in it, I enjoyed writing it, and maybe it might spark a thought for you. Around how you might use AI in your own life. We are already deciding what we want to let AI do for us, and of course the possibilities are many. But I wonder if the bravest choice for you and I is what we won’t let AI do. Would you let it write your mother’s obituary? A speech for your best friend’s wedding? Will you let it make a decision for you about who you should love? How you should live? As a professional, will you hand the client the 12-second version it spat out for you, or will you lean in and decide to present your best work?

There’s no right or wrong answer, I just have this growing feeling that the smartest and most creative people will use it very deliberately.